What would you do if I told you that there are hidden cycles that all living and non-living things follow, cycles that recur as consistently as the four seasons of the year or the waxing and waning of the moon? What if I told you that you could use this hidden knowledge to improve your forecasting skills, so that you could better anticipate price pumps as opposed to being left in the dumps when investing and trading your precious NEO?
In this article, I’ll discuss a tool I created which allows you to look for cycles in the price of NEO. You can access the tool here.
So, how do you use it? Let’s begin.
The tool uses a concept called Time Series Forecasting (TSF), which, given a time series, ‘decomposes’ the data into waves of energy and isolates them into repeating and non-repeating patterns. The repeating pattern is called ‘seasonality’ or ‘cyclicality’, while the the non-repeating patterns are classified as either ‘trend’ or ‘residual’, depending on whether there is a clear pattern or not. Combining these 3 patterns should return you the original, untouched dataset.
TSF relies on the mathematical concept of Fourier analysis, which uses function transforms to discern a clear pattern, or signal, from seemingly random data, or noise. In other words, it takes a time-based pattern, measures every possible cycle, and returns the overall “cycle recipe”.
As an example, Fourier transforms are used in diverse areas such as signal processing (i.e. background noise cancellation), natural language processing (i.e. discerning between different words), and seasonal forecasting. Let’s discuss the last point a bit more closely.
The practice of seasonal decomposition is useful for retailers and e-tailers, because knowing when cycles occur allows them to better plan for times of strong or weak demand. They can better adjust their inventory or prices in advance in order to meet these peaks and troughs, thus maximizing their profit. For example, demand for luxury goods tends to peak around the holidays, especially during the Christmas season, giving retailers incentive to start preparing earlier for the Holidays.
OK, so now that we know what seasonal decomposition is, how can we apply this concept to the price of NEO?
To start, I’d recommend stretching the data set as far back as possible in order to look for patterns. After all, the longer the pattern has been repeating, the more likely it is to recur in the future.
The next step is to enter the period in which you think the cycle or season repeats. Again, period represents the length of the cycle. For example, if you were looking for whether there is an annual period to NEO (i.e. do people tend to buy NEO during Christmas, Easter, etc.), then use 365 days as your period.
Looking at the one-year decomposition, do we see any serious patterns? In fact, we do! There seems to be a mini-peak in the price of NEO that repeats during every April- May period. Those of us who were around in 2017/18 can recall that for some reason, the first big pump, which began in September of 2017, seemed to top out in April of 2018, before fizzling out. This is an interesting finding, since as there isn’t much real-world usage yet, crypto trading really shouldn’t center around specific seasons or holidays. Does anyone have any guesses why late Spring seems to be such a hot time for NEO? If so, please do share your thoughts!
Let’s lengthen the time period. How about 500 days?
OK, that looks interesting too.
There does seem to be another mini-peaking of price that recurs every 500 days or so. Note that this is unlike the 365-day annual cycle, which tends to recur around the same April/ May time frame every year. Like the annual cycle, the mini-peak from bottom-to-top is about $30, which is significant when considering how the price of NEO in recent months.
Let’s try one more cycle, perhaps a bit longer, say 800 days?
Ooooh, we have another nice pattern, but one with a much larger peak-to-trough, at over $50! We can see clearly the recent dump-to-pump that took place from December 2020 to May 2021, when the price of NEO went through its second big acceleration phase.
Now that we’re sort of back in the dumps again, perhaps this chart is telling us when the next pump phase will again occur?
Hmm…
Are we starting to get some ideas here for what the secret cycle might be?
I have my own ideas about what an even bigger pattern might be, but I may save that for a future article.
In the meantime, feel free to use this tool (alongside the NEO regression tool) to see if you can find any patterns yourself. If you like, consider experimenting with both shorter as well as longer periods, to see if you can find a period with an even longer trough-to-peak cycle. (Do note that the length of time of the total dataset must be at least twice as long as the cycle period length, otherwise, the algorithm won’t work since it will be impossible to detect a whether the period is actually recurring.)
Please feel free to sharewhat you’ve found in the comments below.
Thanks, and Happy forecasting!
